NBER WORKING PAPER SERIES
FAMILY PROXIMITY, CHILDCARE, AND WOMEN'S LABOR FORCE ATTACHMENT
Janice ComptonRobert A. Pollak
Working Paper 17678http://www.nber.org/papers/w17678
NATIONAL BUREAU OF ECONOMIC RESEARCH1050 Massachusetts Avenue
Cambridge, MA 02138December 2011
We thank Sarah MacPhee, Karl Scholz, Betsey Stevenson, Christopher Taber, Jeffrey Wooldridge,and participants at the CNISS workshop at Washington University in St. Louis, the Michigan RetirementResearch Center workshop, the NBER Cohort Studies Conference, the University of Wisconsin, theCenter for Retirement Research at Boston College, and the NBER Labor Studies Conference for helpfulcomments. Earlier versions of this paper were presented at the American Economic Association meetingsin Atlanta and the International Association for Time Use Research in Paris. The usual disclaimerapplies. The views expressed herein are those of the authors and do not necessarily reflect the viewsof the National Bureau of Economic Research.
NBER working papers are circulated for discussion and comment purposes. They have not been peer-reviewed or been subject to the review by the NBER Board of Directors that accompanies officialNBER publications.
© 2011 by Janice Compton and Robert A. Pollak. All rights reserved. Short sections of text, not toexceed two paragraphs, may be quoted without explicit permission provided that full credit, including© notice, is given to the source.
Family Proximity, Childcare, and Women's Labor Force AttachmentJanice Compton and Robert A. PollakNBER Working Paper No. 17678December 2011JEL No. J13,J20
ABSTRACT
We show that close geographical proximity to mothers or mothers-in-law has a substantial positiveeffect on the labor supply of married women with young children. We argue that the mechanism throughwhich proximity increases labor supply is the availability of childcare. We interpret availability broadlyenough to include not only regular scheduled childcare during work hours but also an insurance aspectof proximity (e.g., a mother or mother-in-law who can provide irregular or unanticipated childcare).Using two large datasets, the National Survey of Families and Households and the public use filesof the U.S. Census, we find that the predicted probability of employment and labor force participationis 4-10 percentage points higher for married women with young children living in close proximityto their mothers or their mothers-in-law compared with those living further away.
Janice ComptonDepartment of EconomicsUniversity of ManitobaWinnipeg, MB, R3R [email protected]
Robert A. PollakWashington University in St. LouisArts and Sciencesand the Olin Business SchoolCampus Box 11331 Brookings DriveSt. Louis, MO 63130-4899and [email protected]
3
1. Introduction
In this paper we show that close geographical proximity to mothers or mothers-
in-law has a substantial positive effect on the labor supply of married women with young
children.1
Two endogeneity issues complicate estimation. The first complication arises
because childcare decisions and labor supply decisions are often made simultaneously.
We deal with this endogeneity issue by using proximity as an instrument for childcare
transfers from mother or mother-in-law. This IV approach assumes that proximity is
exogenous, an assumption often made in the literature.
Using two large datasets, the National Survey of Families and Households
(NSFH) and the public use files of the U.S. Census, we find that the predicted probability
of employment and labor force participation is 4-10 percentage points higher for married
women with young children living close to their mothers or their mothers-in-law
compared with those living further away. We argue that the availability of childcare is the
mechanism linking proximity and labor supply. We interpret availability broadly enough
to include not only regular scheduled childcare during work hours but also an insurance
aspect of proximity (e.g., a mother or mother-in-law who can provide irregular or
unanticipated childcare).
The second complication arises because of the potential endogeneity of
proximity. If proximity is not exogenous but is related to variables that affect labor
supply and childcare decisions (e.g., both proximity and labor supply may be influenced
by human capital investment, fertility and the marriage market), then the IV results are
misleading. To deal with the potential endogeneity of proximity, we use census data to
analyze the relationship between labor force attachment and proximity using a sample of
“military wives,” civilian women with husbands serving in the U.S. military. Economists
have generally viewed the endogeneity of proximity as less serious than those posed by
the childcare-labor supply relationship, as evidenced by several recent papers that use
proximity to mothers as an instrument for family-provided childcare (e.g., Dimova and
Wolff (2008), Dimova and Wolff (2011), Zamarro (2009)). Estimates based on census
1 We use “mothers” to refer to the older generation, “women” to refer to the middle generation (i.e., the adult daughters of the mothers) and “children” to refer to the youngest generation (i.e., grandchildren of the mothers). By "young children" we mean children 12 and under.
4
data support our conclusion that proximity has a substantial positive effect on the labor
force attachment of married women with young children and that the mechanism is the
availability of childcare.
Using proximity as an instrument for childcare transfers from mothers or
mothers-in-law, we find that married women who receive childcare transfers from their
mothers or mothers-in-law are 5.1 to 6.2 percentage points more likely to work in the
paid labor force than those who do not receive childcare transfers. These estimates are
based on predicted transfer of childcare in the previous month and may underestimate the
full effect of geographic proximity. Proximity to a mother or mother-in-law who can
respond to irregular or unanticipated childcare needs constitutes a kind of insurance the
importance of which may be greater than the number of actual or predicted childcare
hours would suggest. Market-based childcare may be a good substitute for care provided
by a grandmother when the need for childcare is regular and anticipated. But market-
based childcare is less able to meet irregular or unanticipated childcare needs. Hence, the
proximity of a grandmother who can pick up a sick child from school, take a child to after
school sports practice, or care for a child whose parents are traveling on business may
affect women's labor market choices, even if such childcare needs seldom arise.
To allay concerns that proximity affects labor force attachment through
channels other than childcare, we show that proximity has no discernable effect on the
labor force behavior of married women without childcare needs: those without young
children. Although unmarried women with children are more likely than married women
with children to benefit from work-related childcare by their mothers, we do not find a
relationship between proximity and labor force attachment for unmarried women. This
lack of effect is consistent with a more inelastic labor supply of unmarried women with
children, making them less responsive to the availability of childcare.2
Using micro-data from the census, we find additional evidence that proximity to
mothers or mothers-in-law increases the labor force attachment of married women with
2 Kimmel (1998) finds that the labor supply of unmarried mothers is less responsive to childcare prices than the labor supply of married mothers. Bishop et al (2009) estimate labor supply elasticities for single women and show that participation wage elasticities for single mothers have declined dramatically between 1980 and 2004, and are much lower than those for married mothers. They estimate a participation wage elasticity for single mothers of 0.68 in 1979, dropping to 0.25 by 2004. In contrast, Blau and Kahn (2007) estimate wage elasticities for married mothers at 0.98-1.04 in 1979-1981 and 0.48-0.54 in 1999-2001.
5
young children and that the likely mechanism is childcare. Because the census does not
ask about proximity to mothers or mothers-in-law, we use living in one's birth state as a
proxy for proximity. Because the census does not ask about childcare, we estimate the
relationship between labor force attachment and birth state residence. We report three
separate analyses. First, we proceed as we did with the NSFH reduced form equations,
assuming proximity is exogenous and investigating the effect on the labor force
attachment of married women with young children of living in her birth state or her
husband's birth state. Second, we consider a sample of military wives. The military wives
provide an endogeneity control because their husbands' locations are determined by the
military. For the military wives, we find that living in the birth state of both spouses
increases the labor force attachment of married women with young children. Third, we
consider a subsample of migrants -- individuals who, five years prior to the census, were
not living in either their birth state or their current state. We find that married women
with young children who returned to their birth states or to their husbands' birth states
have substantially higher labor force participation than women who moved to a non-birth
state.
Geographical proximity of adult children and their parents has only recently
garnered attention in the economic literature.3
of adult children to their parents as the outcome of a noncooperative game, but they do
Konrad et al. (2002) model the proximity
3 Klerman and Leibowitz (1990) find a non-significant effect of the availability of relative care on the probability of returning to work within 3 months (and also within 24 months) following the birth of a child. Their analyses, however, focus on coresident grandmothers rather than grandmothers in close proximity. Declining rates of coresidence (Costa, 1999; Ruggles, 2007) and the likelihood that coresident grandmothers may themselves need care (Compton and Pollak, 2009) suggest that the focus on coresidence rather than proximity fails to capture the roles of mothers and mothers-in-law. Several recent theoretical papers consider the effect of intergenerational transfers of time on the labor force behavior of daughters. Pezzin and Schone (1999) develop a model in which the labor force participation of daughters and the provision of long-term care to mothers are jointly determined; they focus on the care of frail elderly mothers and do not consider childcare.
6
not consider childcare or labor supply.4 Rainer and Siedler (2009) develop and estimate
a similar model but, unlike Konrad et al., they discuss labor market effects; they find that
adult children without siblings are more likely to remain in their parents’ locations and
have worse labor market outcomes. These findings are strongest when comparing only
children and adults with siblings who grew up in economically depressed regions: the
earnings of adults with siblings are ten percent higher on average, and adults with
siblings are ten percent more likely to be working full time, compared to only children.
They do not investigate the effect of the availability or receipt of childcare on women’s
labor supply.5
Three recent studies consider the effect of childcare by mothers (but not by
mothers-in-law) on the labor force behavior of women in Europe, using proximity as an
instrument for childcare transfers. Using SHARE data, Dimova and Wolff (2011) use a
simultaneous recursive model to estimate the effect of both time and money transfers
from mothers on the labor force participation of their daughters in 10 European
countries.
6
4 In their model, the eldest sibling has the first mover advantage and moves away from the parents to shift the burden of providing long-term care for elderly parents to younger siblings.
They include distance between mothers and daughters as well as mothers’
demographic characteristics in their childcare equation. They find that regular (weekly or
daily) transfers of childcare have a small positive effect on daughter’s labor force
participation, but do not affect whether their labor force participation is full-time or part-
time. Using the same data and a recursive simultaneous equations model, Zamarro (2009)
considers the country-specific impact of regular childcare transfers on the labor supply of
both mothers and daughters. She finds that regular childcare transfers affect the
daughters' labor supply for Greece and the Netherlands, but are insignificant for the other
8 countries. Finally, using French data Dimova and Wolff (2008) find that daughters of
first-generation immigrant women at or near retirement age are more likely to participate
in the labor force if they receive regular (i.e., weekly) childcare from their mothers.
5 Cardia and Ng (2003) calibrate an overlapping generations model that allows intergenerational transfers of both time and money; they show that time transfers involving childcare have substantial positive effects on the labor supply of the middle generation. Belan, Messe and Wolff (2009) develop and analyze an overlapping generations model with intergenerational transfers of care and show that changes in the mandatory retirement age affect the employment rates of both generations. 6 SHARE, the Survey of Health, Ageing and Retirement in Europe, is a large multi-country panel covering more than 45,000 individuals over the age of 50.
7
Arpino, Pronzato and Tavares (2011) do not consider proximity, but estimate
the effect of grandparent-provided childcare on the labor force attachment of women in
Italy using the number of living grandparents as an instrument for childcare transfers.
They find a very large (30 percentage point) positive effect of grandparent care on the
probability that a woman is working. Finally, Compton (2011) estimates the direct effect
of proximity to mother on the labor force attachment of women in Canada. She finds a
substantial positive effect of proximity on the probability of work (11 percentage points),
only for married women with young children.
We argue that the empirical strategy used in the majority of these studies
underestimates the effect of childcare availability for two reasons. First, with the
exception of Compton (2011), they focus on regular childcare, ignoring the insurance
aspect of childcare to meet irregular or unanticipated needs. Second, with the exception
of Arpino, Pronzato and Tavares (2011), they consider only childcare transfers from
mothers to daughters because the data sets they use do not include information on
mothers-in-law. We find that the effect of proximity on labor force attachment is
strongest for those women living near both mothers (in the NSFH data) or living in the
birth state of both spouses (in the census data). "Living near neither mother" has a strong
negative effect on labor force attachment. The effect of close proximity to only her
mother or only his mother is positive, but not robust across samples. In the NSFH, the
effect of close proximity to only his mother is positive and significant, while the effect of
close proximity to only her mother is insignificant. In the census data, living in the birth
state of either spouse has a significant positive effect on labor force attachment and the
magnitudes are about the same. For the military wives sample, we find significant effects
on labor force attachment only for couples that live in the birth states of both spouses.
Thus, having data on proximity to both mothers and mothers-in-law or on the birth states
of both spouses is crucial to understanding the effect of proximity on labor force
attachment.
Our finding that family proximity increases the labor force attachment and
employment of married women with young children has implications for policy. The
magnitude of the effect of proximity, 5.2 to 10.4 percentage points, is similar to that
found for MSA status, race. and ethnicity. For example, our results suggest that married
8
women are 3.8 percentage points more likely to work if they live in an MSA, 6.9
percentage points more likely to work if they are Black, and 8.2 percentage points less
likely to work if they are Hispanic. The magnitudes are also similar to those found for
education: women are 6.0 percentage points more likely to work if they are in a power
couple (both spouses have a college degree), 15 percentage points more likely to work if
they have a college degree but their husband does not, compared to a woman in a low-
power couple (neither spouse has a college degree).7
This paper proceeds as follows. Section two describes the NSFH and uses it to
analyze the relationship between childcare, proximity and women's labor force
attachment. Results from IV and reduced form probit regressions on the probability of
being employed support the hypothesis that proximity to mothers or mothers-in-law has a
substantial positive effect on the labor force attachment of married women with young
children, but not on that of any other demographic group. Tobit and selection correction
models on hours of work tell a similar story. We conclude section two by discussing three
alternatives to our childcare explanation of the positive association between proximity
and the labor force attachment of married women with young children: differences in
husbands’ incomes, informal job-search networks, and extended childcare networks. We
discuss a fourth alternative explanation, tied mover effects, at the end of section three
Our analysis suggests that policies
that increase the availability of childcare to meet irregular or unanticipated child care
needs, including care for sick children, might substantially increase the labor supply of
married women with young children. Our analysis also suggests that increases in the
retirement age which reduce the ability of the older generation to provide childcare may
reduce the labor force attachment of daughters in the younger generation. Discussing
recent trends in labor force participation in the U.S., Mosisa and Hipple (2006) note that
while participation rates have decreased in the past decade for women aged 25 to 54, they
have increased for women aged 55 and older. The behavior of these cohorts is usually
analyzed separately, with little or no recognition that geographical proximity and
childcare may provide a link between them.
In section three we turn to census data. Because the census does not ask about
proximity to mothers or mothers-in-law, as a proxy we investigate the effect of living in
7 We have borrowed the "power couples" terminology from Costa and Kahn (2000).
9
one’s birth state. The results for married women with young children and for military
wives reinforce our conclusion that proximity to mothers or mothers-in-law increases the
labor force attachment of married women with young children. We also compare the
labor force attachment of a subsample of recent migrants: individuals who, five years
before the census, were not living in either their birth state or their current state.
Comparing those who returned to their birth states with those who moved on to another
state, we find that those who returned to their birth states have higher labor force
attachment. This finding provides additional support for our inference that proximity
increases the labor force attachment of married women with young children. In section
four we summarize our findings and conclude.
2. NSFH: Proximity and Labor Force Attachment
We use data from the first two waves of the National Survey of Families and
Households (NSFH) described in Sweet and Bumpass (1996). The first wave (1987-
1988) consisted of 13,007 households, and oversampled blacks, Puerto Ricans, Mexican
Americans, single-parent families, families with stepchildren, cohabiting couples, and
recently married couples. The second wave (1992-1994) was a five-year follow-up.
Using the first and second waves of the NSFH enables us to control for recent migration
(i.e., living in a different location in the second wave than in the first.).8, 9
respondent was randomly chosen from the adults in the household, but both the
respondent and the respondent's spouse or partner were asked to complete the entire
survey. The data include information on distance (in miles), health, marital status,
education, and transfers given and received by both the respondent’s mother and mother-
in-law.
The primary
10
60 and whose mothers (and mothers-in-law where applicable) are Alive and Living in the
United States (ALUS). Thus, we exclude individuals whose mothers or mothers-in-law
We limit our sample to those women (respondent or spouse) who are aged 25
to
8 The NSFH does not provide information that would enable us to control for migration prior to wave 1. 9 The third wave of the NSFH sample was reduced to include only households with children. Because this sample restriction limits our ability to compare across groups, we use only the first and second waves. 10 We use the information collected from the respondent; if this information is missing, we use the spouse’s record.
10
are deceased or live outside the U.S.11
For our analysis, a major advantage of the NSFH
is that it provides information on proximity not only to mothers but also to mothers-in-
law. Although few data sets include information about family proximity, our results
suggest a high scientific payoff to collecting and analyzing information about proximity
to both mothers and mothers-in-law.
2.1. Description
Most Americans live very close to their mothers. Using data from the NSFH,
Compton and Pollak (2009) report that the median distance between married women and
their mothers is 20 miles, with one-quarter living within 5 miles of their mothers.
Unmarried women live even closer: the median distance is 8 miles when coresidents are
included in the distance calculation and 15 miles when they are excluded.12 We define
‘close proximity’ or ‘living near’ as a distance of twenty-five miles or less.13
The NSFH provides information on time transfers between individuals and their
mothers and mothers-in-law. Respondents were asked whether, in the previous month,
they provided or received general help (shopping, errands, transportation, housework,
yard work, car repairs and other help around the house) to or from their parents or
parents-in-law. Those with children 12 and under were asked whether they received
childcare from their parents or parents-in-law while working or childcare at other times
Close
proximity is strongly correlated with education: 46 percent of low-power couples
(couples in which neither spouse has a college degree) live within 25 miles of both
mothers, whereas only 17 percent of power couples (couples in which both spouses have
college degrees) live within 25 miles of both mothers.
11 Although the data are fifteen years old, patterns of migration and proximity appear to be quite stable. For example, the percentages of individuals living in their birth state (our proxy for proximity when using census data) has remained fairly constant over the past three decades. 12 In the NSFH analysis, unmarried women include those who are never married, divorced, widowed or separated. We include cohabitors with married individuals. 13 The results of the analyses are very similar if cutoffs of 20 miles or 30 miles are used. Unless noted otherwise, we include couples who coreside with either her mother or his mother in the ‘close’ category. Although these couples are qualitatively different from those not coresiding (see Compton and Pollak, 2009), they are a small proportion of the population (2.4 percent of the sample) and sample sizes are too small to justify a separate category. If we exclude coresidents from the sample of married women, the results are indistinguishable. We treat unmarried women who coreside with their mothers as a separate category because the sample size is larger; 22 percent of unmarried women live with their mothers.
11
(table 1). The likelihood of time transfers is strongly associated with proximity to both
mother and mother-in-law. Of married women with young children living within 25 miles
of their mothers, 24-27 percent received work-related childcare while 31-37 percent
received non-work-related childcare from their mothers; of married women with young
children living within 25 miles of their mothers-in-law, 18-19 percent received work-
related childcare and 25 percent received non-work-related childcare from their mothers-
in-law. Unmarried women with young children were slightly more likely to receive
transfers of work-related childcare from her mother: 28 percent of unmarried women
young in close proximity to their mothers received work-related childcare and almost
one-half received non-work related childcare in the past month. Those living further than
25 miles were much less likely to receive childcare: only 4.2 percent (2.7 percent) of
married women with young children who did not live close to either mother received
work-related childcare from her (his) mother.
Employment is also correlated with proximity. Table 2 shows the labor force
attachment of married and unmarried women by proximity to their mothers or mothers-
in-law. For unmarried women there is a positive relationship between distance category
(coresidence, 25 miles or less, more than 25 miles) and full-time work, but an inverted U-
shaped relationship between distance category and out of the workforce (the sample size
here is a concern, however). This pattern is observed both for unmarried women with
young children, and those without young children. For married women, there are four
categories of proximity: a couple can live close to neither mother, to his mother only, to
her mother only, or to both mothers. The raw data show the importance of including both
mothers and mothers-in-law when considering the relationship between proximity and
labor force attachment. If we exclude information on mothers-in-law, we are in effect
combining the first two categories into a single category ("not close to her mother") and
the last two categories into a single category ("close to her mother"). Yet, married
women, especially those with children, who live near only their mothers-in-law have a
much different pattern of labor force attachment than married women who do not live
near either mother. Married women who live near only their mothers have a much
different pattern of labor force attachment than married women who live near both
mothers. For example, restricting our attention to married women with young children,
12
we find a substantially higher percentage working full-time when living near only their
mothers-in-law (45 percent) than living near neither mother (33 percent). By recognizing
four proximity categories, we are able to estimate more precisely the effect of proximity
on labor force attachment.
Demographic factors correlated with close proximity are typically factors
correlated with lower labor force attachment. Means and standard deviations for the
married women's sample are presented in Appendix 1. Compared with women who do
not live within 25 miles of either mother or mother-in-law, those who live close to both
live in areas of higher unemployment, are younger, are more likely to have young
children, are less educated, have less educated mothers and spouses, are more likely to be
black or Hispanic, and are less likely to live in an MSA. Yet despite these correlates of
close proximity, women living in close proximity to their mothers or mothers-in-law are
more likely to be working and work more hours. In the next sub-section, we show that the
proximity effect observed in the raw data holds under regression analysis.
2.2. Analysis: Childcare and Labor Force Attachment
We begin by estimating the effect of predicted transfers of childcare on the labor
force behavior of adult women, similar to the type of analysis performed on the European
data by Dimova and Wolf (2008, 2011) and by Zamarro (2009). The sample is restricted
to women with young children whose mothers are ALUS.14
2,1,0001
)2(
)1(
*
*
21*
12*
2
12*
2*
1
=≤=>=
++Λ+Χ=
++Ζ+Χ=
kYifYYifY
UYY
UYY
kiki
kiki
iiiii
iiiii
ϕγβ
ϕλβ
We estimate the impact of
both work-related and non-work related childcare, using the following bivariate probit
model:
14 Women coresiding with their mothers are excluded from the childcare regression samples as transfers between coresidents are not included in the data.
13
where iY1 is an observed dichotomous variable equal to 1 if the daughter works positive
hours, (i.e., if the latent variable *1iY >0); iY2 is an observed dichotomous variable equal to
1 if the daughter receives childcare (either work related or non-work related) from her
mother or her mother-in-law (i.e., if the latent variable *2iY >0). The vector X includes
exogenous control variables common to both regressions (age, age squared, whether
husband works and his hours of work, husband’s income, whether self or husband
currently has medical problems, race (black, Hispanic, white (omitted)), education
categories (both spouses have college degrees, only she has a college degree, only he has
a college degree, neither has a college degree (omitted)), age of youngest child, whether
mother has a college degree). The vector Z includes the instruments for work (region
(Midwest, South, West, Northeast (omitted)), average commuting time in the county (to
account for place-to-place differences in the amount of time it takes to travel), whether
residing in an MSA, 1990 county level unemployment rate, and whether the respondent
lived in a different city in the first wave of the data). The vector Λ includes the
instruments for childcare (age categories of mother(s) (less than 60, 60-69, 70 and over
(omitted)), whether mother(s) are in poor health, whether mother(s) are married and
whether mother(s) live in close proximity). Error terms are assumed to be iid normal. The
variables included in Λ are assumed to affect the likelihood of childcare, but not labor
market behavior directly, while variables included in Z are assumed to affect the
likelihood of working, but not childcare. The model is estimated using the two-step
procedure outlined in Maddala (1983) and Greene (1998).15
The parameter of interest is
2ϕ , the coefficient on predicted childcare in equation
(1). This is presented in Table 3.16
15 While this procedure gives consistent estimators, it is not efficient. We use the econometrics software Stata for this analysis, and this software does not allow for a simultaneous bivariate probit estimator with two endogenous variables. Standard errors are estimated using a bootstrap procedure.
We analyze the relationship between proximity and
childcare using three definitions of childcare – work-related childcare, non-work-related
childcare, and either type of childcare. In the top panel, we show the results for unmarried
women. For this sample, transfers of childcare have no discernable effect on the
16 In all regressions, coefficients are presented with standard errors in parentheses. Marginal effects on the predicted probability are shown in italics.
14
probability of work. For the sample of married women, we first estimate the effect of
transfers to and from her mother (ignoring mother-in-law information). In these
regressions, the variables inΛ include only those that pertain to her mother, including
whether the woman lives in close proximity to her mother. The results show positive, but
insignificant coefficients on predicted childcare. The bottom panel adds the information
on mother-in-law. 17 Λ Here, the variables in include the age, marital status and health of
both mothers and the proximity categories are close proximity to his mother, close
proximity to her mother and close proximity to both. The results show that married
women who receive transfers of childcare from either their mother or mother-in-law are
more likely to work in the paid labor force than those who do not receive such transfers.
Marginal effects are significant, ranging from 5.1 to 6.2 percentage points, depending on
the definition of childcare. The effect is significant for both work-related and non-work
related childcare.18
We next investigate the reduced form relationship between proximity to mothers
and mothers-in-law and the labor force behavior of women, maintaining the assumption
that proximity is exogenous. We highlight the reduced form analysis for two reasons.
First, the childcare variable in the NSFH indicates only whether a woman received
childcare from her mother or mother-in-law in the previous month. The IV analysis will
underestimate the insurance aspect of childcare if the mother or mother-in-law is
available for childcare, but no transfers were received during the survey month. Second,
applying the reduced form equation to demographic groups without childcare needs
We interpret the significance of non-work-related transfers as
suggesting an insurance effect of proximity – the availability of family members to
provide childcare in unanticipated situations alters the labor supply of married women.
17 There is no econometric procedure to allow for a two-step IV approach with a multinomial endogenous variable (childcare from her mother only, from his mother only, from both). 18 The categories of childcare (columns A and B) are not mutually exclusive. For unmarried women, 63 percent of those who received work-related childcare also received non-work related childcare; 40 percent of those who received non-work related childcare also received work-related childcare. For married women, 77 percent of those who received work-related childcare from either mother or mother-in-law also received non-work related childcare; 58 percent of those who received non-work related childcare from either mother or mother-in-law also received work related childcare.
15
allows us to test our hypothesis that the availability of childcare is the mechanism that
links proximity and labor force attachment.
Using the notation above, we estimate
)3(*1 iiiii UY +Λ+Ζ+Χ= γλβ
as a reduced form probit regression on employment. We also estimate Tobit regressions
with iY1 denoting hours of work.
We now expand our sample to include all women, not only those with children
under the age of 12. For married women, we focus on the effect of three categorical
variables: close proximity (i.e., within 25 miles) of mothers only, of mothers-in-law only,
and of both mothers and mothers-in-law. For unmarried women, we consider the effect of
coresidence and of close proximity to mothers. To simplify the interpretation, we limit
the sample to those with mothers (and mothers-in-law for the married sample) ALUS.19
Control variables indicating the presence of children (children 12 and under, only
children older than 12, no children (omitted)) are added to vector X.20
Probit results are shown in table 4. In columns (1) and (3) we estimate the effect
of living near own mother, ignoring the location of mother-in-law. As with the IV
regressions, we find positive but insignificant effects of proximity when the comparison
group contains both those living near neither mother and those living near their mother-
in-law only. When mother-in-law information is added in columns (2) and (4), the
comparison group becomes those living away from both mothers and we now see a
statistically significant and relatively strong effect of proximity to mothers-in-law and to
both mothers. For unmarried women with young children, we find no effect of proximity,
and a negative effect of coresidence, on work force attachment. We replicated these
regressions for married and unmarried men with young children and found no significant
effect of close proximity on men's labor force attachment.
19 By excluding those whose mothers are not ALUS, our sample under-represents migrants to the U.S. and those whose mothers die young. 20 These are mutually exclusive categories. The category “only children older than 12” includes women with older children at home, and those who have older children living elsewhere. These groups are combined for sample size. The category “no children” includes only those women who have never had children.
16
In table 5 we consider different subsamples of married women to determine the
subgroups for which the relationship between proximity and labor supply is strongest. 21
The first column presents the full sample results from the regression in table 4, column 2.
In columns (2) – (4), we report the regressions separately by presence of child categories:
column (2) includes only those with young children; column (3) includes only mothers
without young children; and column (4) includes only non-mothers.22
In columns (5) and (6) we limit the sample to those whose mothers or mothers-in-
law are in poor health and thus are more likely to need care themselves and less likely to
provide care for their grandchildren. We find no effect of proximity on the labor force
attachment of these women. The absence of an effect of proximity on the labor supply of
women whose mothers or mothers-in-law are in poor health is further evidence that the
availability of childcare is probably the mechanism through which proximity affects labor
supply.
Proximity is
significant only for those with young children and the effect is large; close proximity to
mother-in-law or to both mother and mother-in-law increases the predicted probability of
employment by 10 percentage points. The coefficient on close proximity to only her
mother is positive, but insignificant.
Proximity to mother-in-law and proximity to both mothers have similar effects:
proximity to only one’s own mother has a smaller and statistically insignificant effect.
This result is unexpected, as women are more likely to receive childcare transfers from
their mothers than from their mothers-in-law. In section 5, using census data and birth
state residence as a proxy for proximity, we find that proximity to mother and proximity
to mother-in-law are statistically significant and that the effect sizes are about the same.
The effect of nearby siblings suggests that strategic behavior may explain why
proximity to mother-in-law has a stronger estimated effect than proximity to mother.
Consider first the mother-in-law. Because mothers-in-law are more likely to provide
21 We found similar results when we included interaction terms between proximity and children or health of mother or mother-in-law in the regression. 22 The results are qualitatively the same if we consider those with children under the age of 6. We chose the 12 year old cut-off for two reasons. First, this cut-off corresponds to the NSFH childcare transfer questions – only those with children 12 and under were asked about childcare. Second, our hypothesis is that the availability of family to aid with irregular or unanticipated childcare needs is important for labor market decisions. This type of childcare may be especially important when children are school-age.
17
childcare for the children of their own daughters than for those of their daughters-in-law,
the presence of the mothers-in-law's other children may reduce her willingness to provide
childcare. On the other hand, if there are no siblings in close proximity, mothers-in-law
may have a stronger incentive to provide childcare transfers than mothers. Because
altruistic motives for providing eldercare are presumably weaker among daughters-in-law
than daughters, mothers-in-law may be more willing to provide childcare to daughters-in-
law in the hope of increasing the probability of receiving eldercare in the future. In the
first column we find a positive effect of living close to one’s mother-in-law but a
negative effect of close proximity to husband's siblings.23
The results on hours of work from Tobit regressions and models using a Heckman
correction for sample selection indicate that the effect of proximity is primarily on the
extensive margin (i.e., whether the woman works or not) rather than on the intensive
margin (i.e., the number of hours worked). These results (not shown) are consistent with
the probit results: we find proximity effects only for married women with young children.
The negative effect of nearby
siblings is also seen for the subgroup with young children, although these coefficients just
fail to meet standard levels of significance. Strategic behavior may also explain the
insignificant effect of close proximity to only her mother. Compared with couples
residing in close proximity to both mothers, those residing in close proximity to only her
mother may be more likely to move away in the future, thus reducing the incentives of
mothers to provide childcare. The close proximity of a woman’s own siblings has a
negative but non-significant effect on the labor force attachment of women with young
children, and a positive effect on the labor force attachment of those without children.
These results suggest a relationship between labor force attachment and sibling
competition in care transfers, but we do not have sufficient data to investigate this
possibility more thoroughly.
We consider four alternatives to the childcare explanations of the positive
association between proximity and the labor force attachment of married women with
young children: differences in husbands’ incomes, informal job-search networks,
extended childcare networks, and tied mover effects. We discuss husbands’ incomes,
23 We do not know whether siblings in close proximity are brothers or sisters.
18
informal job search networks, and extended child care networks in this section and tied
mover effects in section 3.2
We discount the husbands' incomes as an explanation of our proximity results
because we have included husbands’ incomes, hours of work, work status and education
as control variables in all of our regressions. The raw data show that married women
who live in close proximity to their mothers or mothers-in-law have husbands with lower
incomes than those who live far from both. Husbands are also more likely to work and
work more hours when the couple lives far from both mothers, compared to those living
close to either mother or mother-in-law. A joint model of household labor force
attachment would predict, based on this information, that married women living away
from both their mothers and their mothers-in-law would have lower labor force
attachment than those living close. But the fact that this relationship persists after
controlling for husbands' incomes, hours of work, work status and education makes this
an unlikely alternative to the child care explanation.
We discount the informal job-search networks explanation of our proximity
results because we find an effect of proximity only for married women with young
children. We do not find a proximity effect for men, for unmarried women, or for married
women without young children. Even if we postulate that the labor force attachment of
unmarried women with young children is more sensitive than that of other demographic
to the presence of a more extensive job search network, we would expect to find some job
search network effect for other groups. The absence of such effects argues against the job
search network explanation.
We are skeptical of the extended child care network explanation of our proximity
results, but we cannot rule out the possibility that proximity to mother and mother-in-law
is a proxy for extended childcare networks involving siblings, aunts, uncles, cousins, and
friends. Three points need to be made. First, after controlling for a woman’s proximity to
her mother, proximity to her siblings has no effect on her labor force attachment, while
proximity to her husband’s siblings has a negative effect. This suggests that there is no
additional increase in labor force attachment associated with living in close proximity to
other family members. Second, the connection between proximity and labor force
attachment is not observed when mothers are in poor health. Finally, using data from the
19
Survey of Income and Program Population (SIPP), Laughlin (2010) shows that
grandparent-provided childcare is three times more prevalent than care provided by other
relatives -- that is, grandparents provide a disproportionate share of nonparental
childcare. However, the "insurance" story (e.g., having someone to pick up a sick child at
school) could be told about siblings, aunts, uncles, cousins and friends, and we have
argued that the importance of such insurance is not necessarily reflected in the number of
hours of child care provided. Hence, although we are skeptical of the extended child care
network story, we cannot rule it out.
Overall, our findings from the NSFH indicate that proximity to mothers or
mothers-in-law has a large positive effect on the labor force attachment of married
women with young children. The IV analysis suggests that the mechanism through which
proximity and labor force attachment are linked is childcare; the lack of a proximity
effect for married women without childcare needs reinforces this result. Our estimation
of the marginal effect of the availability of childcare is higher than the European results
of Dimova and Wolff (2008, 2011) and Zamarro (2009). We attribute this to our ability
to distinguish both proximity to mother and proximity to mother-in-law, and the focus of
the European studies on regular childcare transfers. Our marginal effects are just slightly
smaller than the Canadian findings (Compton, 2011).24
Two caveats are required. The first is sample size: perhaps the insignificant
results for married women with older children and for married women with no children
are due to the small sub-samples. The second is the potential endogeneity of proximity.
Endogeneity problems arise if women who have preferences for both children and labor
force attachment are more likely to reside near family, compared with women who have
preferences for one or the other.
25
24 Although the Canadian study uses a similar methodology in focusing on proximity rather than childcare transfers, mother-in-law location is not available in the Canadian data. As noted earlier, this exclusion is likely to cause an underestimation of the proximity effect. Still, Compton (2011) finds that married women with children who live in close proximity to their mothers are 11 percentage points more likely to work than those who live more than half a day away.
Unfortunately, we have no convincing way to deal
with
25 Endogeneity problems also arise if marriage market choices reflect underlying preferences for work/children combinations.
20
this endogeneity problem using the NSFH data.26
Nevertheless, our empirical results
from the NSFH provide strong evidence that proximity is related to the labor force
attachment of married women with young children, and that the mechanism is the
availability of childcare. In section three we address the sample size and endogeneity
concerns using census data.
3. Census Data: Birth State and Labor Force Attachment
Although the U.S. Census does not ask respondents the distance to their mothers,
it does ask whether the respondent resides in his or her birth state. We use this variable as
a proxy for close proximity. Data from the Panel Survey of Income Dynamics (PSID)
provides some support for the validity of this proxy.27 The PSID reports grouped
distance to mother in one year (1988) and the “State where the Head (Spouse) grew
up.”28
Using census data we estimate the effect of birth state residence on the probability
of employment and labor force participation as well as on usual weekly hours. More
specifically, using the 2000 public use microdata files of the 2000 U.S. Census, we
construct a dataset that includes all women aged 25-45 who were born in the U.S.
Although the state where one grows up need not coincide with birth state, there is
a strong link between proximity and residing in one's childhood state: in the PSID, more
than 90 percent of heads currently living in their childhood state are living in the same
state as their mothers; over half live within 10 miles, and less than 15 percent live more
than 100 miles away. On the other hand, of those heads not living in their childhood state,
only 27 percent currently live in the same state as their mothers; 16 percent live within 10
miles, and more than 70 percent live more than 100 miles away.
26 We attempted a bivariate probit model as outlined above in equations (1) and (2) , but defining iY2 as an observed dichotomous variable equal to 1 if the daughter lives in close proximity to his mother or to her mother (i.e., if the latent variable *
2iY >0). The instruments included were mother’s marital status and indicators for only and eldest child. We do not present the results for two reasons. First, the results are insignificant and sensitive to control inclusion, which may reflect the use of a binary proximity category that ignores the location of mothers-in-law. Second, although mother’s marital status and birth order are strong predictors of proximity in previous work (e.g., Konrad et al. (2002), Rainer and Siedler (2009), Compton and Pollak (2009)), we found them to be borderline weak instruments, especially in sub-samples. 27 The NSFH does not include state of birth. 28 The PSID did not ask state of birth until 1993, and then only to new heads or spouses. The distance groups in the PSID are (1) less than one mile; (2) 1 to 10 miles; (3) 11 to 100 miles; and (4) more than 100 miles.
21
(Ruggles et al., 2009). For married women, we define three mutually exclusive indicator
variables: (1) whether the couple lives in the birth state of both spouses; (2) whether the
couple lives in only her birth state and (3) whether the couple lives in only his birth state.
To control for migration effects, we include a dummy variable for whether the woman
was in the same state five years previously. We also include controls for the geographic
size of the current state; we do this because those living in large birth states (e.g., Texas,
California) may well have moved within the state and, hence, living in a large birth state
is likely to be a weaker proxy for proximity to mother than living in a small birth state
(e.g., Rhode Island, Delaware). In contrast, those living in small birth states, even if they
have moved within the state, are more likely to live in close proximity to mother.
Because the census provides no information on mothers who do not reside with their
adult children, we limit the sample to those aged 25-45 (in the NSFH analysis we used
those aged 25-60) to increase the likelihood that the mothers of those in our census
sample are still alive.29
We replicate the analysis using a large sample of military wives – civilian women
with husbands serving in the U.S. military.
30 Our military wives sample includes 14,833
married women, of whom 10.2 percent live in only her birth state, 5.1 percent live in only
his birth state, and 8.7 percent live in the birth state of both spouses. The military wives
provide a control for endogeneity because their husbands' locations are determined by the
military.31
3.1. Interaction of Birth State and Young Children
We consider the impact of birth state residence for three samples – married
29 Summary statistics for the Census data are presented in Appendices 2 and 3. 30 Excluded from the sample are those for whom spouse is absent. In particular, this excludes military wives whose husbands are serving overseas. To increase the size of the military wives sample, we have increased the age range to 18-45. We find similar results when we omit the 18-23 age category. 31 The probability that they live near only her mother is twice the probability of living near his mother, presumably because it is the husbands' locations that are determined by the military. In particular, if the wife met her husband while he was in the military and stationed in her birth state, then the couple is more likely to live near her mother than near his mother. Unfortunately, the census does not ask how long couples have been married.
22
women, military wives, and never-married women.32
In table 7 we present regressions analogous to equation (3).
Table 6 presents summary statistics
for the samples. The data indicate that, for married women with young children, there is
an increasing attachment to the work force as we move from residing in the birth state of
neither spouse to residing in the birth state of both spouses. We find no discernable
pattern for married women with only older children or no children. We find a similar
pattern for military wives with young children, although the patterns for military wives
with older or no children are less clear. For never married women with young children,
we find a negative relationship between birth state residence and labor force attachment. 33 That is, we estimate
the effect of birth state residence (our proxy for family proximity) on labor force
attachment – whether the woman is currently in the labor force and whether she is
currently employed.34
The results from the military sample are weaker: the results in table 7 indicate a
positive effect of birth state residence for those with young children, although the
For married women with no young children, we find a small
negative effect of birth state residence, but for married women with young children, we
find a positive effect. Proximity has a small, negative effect on the labor force attachment
of never married women regardless of whether they have young children. The marginal
effects are smaller in the census than in the NSFH sample but the effect remains
substantial: birth state residence increases the probability of labor force participation and
employment of married women with young children by 2.6 – 3.9 percentage points. In
table 8, we expand the birth state categories to account for residence in his and her birth
state separately. Results for the full sample of married women are consistent – a small,
negative effect of living in the birth state of one or both spouses for those without young
children, but a strong positive effect of living in the birth state of either or both spouses
for those with young children.
32 In the NSFH sample we included all unmarried women with controls for divorced, separated and widowed. With the large sample size available in the IPUMS data, we are able to consider separately those previously married and those never married. The results for those previously married are more difficult to interpret since the women may still reside near their mothers-in-law and receive childcare from them. 33 Due to computing demand, a random 10 percent sample was drawn for the regressions. 34 Regression results from Tobit and Heckman corrected models on usual weekly hours again suggest that the impact of proximity is on the extensive margin. The results of these regressions are not presented but are consistent with the probit results.
23
significance levels are relatively low.35
In table 8, with expanded birth state categories,
the interaction between children and birth state residence is positive and significant only
for those residing in the birth state of both spouses. We expect weaker results in the
military wives sample for two reasons. First, birth state residence is a weaker proxy for
family proximity when the husband is in the military because military personnel assigned
to their birth states are likely to be further from their mothers than civilians who live in
their birth states. Second, the strategic motivation for mothers and mothers-in-law to
provide childcare in anticipation of reciprocity when they are elderly and disabled is
reduced because daughters and daughters-in-law are likely to move when their husbands
are assigned to a different location.
3.2. Migration: Origin/ Destination Effects
In this section we investigation origin/ destination effects for migrants. We
consider a subsample of "recent migrants" by which we mean individuals who, five years
prior to the census, were not living in either their birth state or their current state. We
find that those who returned to their birth states have substantially higher labor force
attachment than those who moved to another state.
Although the "tied mover" hypothesis described by Sjaastad (1962) Mincer
(1978), Lichter (1983) and Greenwood (1985) does not explain our proximity results, our
results imply the need to disentangle proximity effects and tied mover effects. The tied
mover hypothesis postulates that the costs of migration are higher if both spouses are
attached to the labor force, and concludes that single-earner couples are more likely to
migrate than two-earner couples. The tied mover hypothesis implies that secondary
earners (read: married women) who migrate will have less labor force attachment, at least
in the short run, than secondary earners who do not migrate.
In table 9 we show the employment and labor force attachment rates for
migrants and non-migrants. We limit our sample of migrants to those who were not living
in their birth state five years earlier because we want to distinguish between the labor
force attachment of those who returned to their birth states (return migrants) and those
35 For the labor force and employment participation probits, the coefficients on the interaction term are significant at the 89 percent and 78 percent confidence levels, respectively.
24
who moved to a state other than their birth state (onward migrants).36
In table 10 we use regression analysis to further investigate these origin/
destination effects. We present the results for probit regressions on labor force
participation of women who did not reside in their birth state five years prior (regressions
on employment and hours yield similar results). The results confirm the patterns observed
in the raw data: labor force participation of married women is negatively related to
migration, but destination is also important. For married women with young children, the
negative effect of migration on labor force participation is substantially less for those
who move back to their birth state (i.e., return migrants) than for those who move to a
different another state (i.e., onward migrants). For married women with no children or
those with only older children, the tied mover effect is smaller and there is no discernable
difference between the two migration coefficients: the effect of returning to one's birth
state is the same as the effect of moving elsewhere.
For married
women with young children, non-migrants have higher labor force attachment rates and
higher employment rates than migrants, which is consistent with the tied mover
hypothesis. Within the group of migrants, however, the participation and employment
rates of married women with young children who return to their birth state are 5.5
percentage points higher than that of their counterparts who migrate to another state. For
the other samples – unmarried women, married women with no children, and married
women with only older children – return migrants have lower labor force participation
and employment rates than onward migrants.
Five points about the relationship between the tied mover hypothesis and our
proximity results deserve attention. First, the tied mover hypothesis, as its name suggests,
applies only to those who moved as a couple; our analysis, on the other hand, focuses on
the proximity of a couple to his mother or her mother, regardless of whether they moved
36 We cannot identify individuals who moved between states within the five year window and then returned, nor can we identify those who moved within state.
25
as a couple or as unmarried individuals.37,38
These origin/destination results imply that the tied mover hypothesis cannot
explain the proximity effects that we have found. They also imply that tied mover effects
and proximity effects interact: the effect of migration on labor force attachment depends
on the presence or absence of young children and on the destination of the migrant.
Second, we find a positive effect of
proximity only for married women with young children, while the tied mover hypothesis
applies to all secondary earners who migrated as part of a couple. Third, we include
controls for recent migration in all regressions. Although this does not capture long-run
effects of migration, a number of studies indicate that the disruptions to wives’ labor
force participation are relatively short-lived (e.g., Clark and Withers (2002), LeClere and
McLaughlin (1997), Marr and Millerd (1988), Spitze (1984)). Fourth, for married women
with young children, we find a positive effect of close proximity to mothers-in-law even
for those women who do not live in close proximity to their own mothers. Because
women living near their mothers-in-law but not near their own mothers are more likely to
be tied movers than those living near their own mothers but not their mothers-in-law, the
tied mover hypothesis predicts a more negative effect of close proximity to mothers-in-
law only. This is not what we find in NSFH or in the census. Finally, using census data,
we find that for married women with young children, the effect of moving back to their
birth state has a less negative effect on labor force participation than moving elsewhere.
4. Conclusion Using two large U.S. data sets, the NSFH and the census, we find that living close
to mothers or mothers-in-law has a strong positive effect on the labor force attachment of
married women with young children. More specifically, we find that proximity increases
the labor force attachment of married women with young children by 4-10 percentage
37 The census does not provide information on whether they moved as unmarried individuals or as a couple. Many migrants are never married individuals. The 2001 Current Population Survey data show that while never married individuals comprise 28 percent of the population over the age of 15, 40 percent of inter-county migrants and 41 percent of inter-state migrants are never married. (Calculations by authors from the 2001 Current Population Survey data found http://www.census.gov/population/www/socdemo/migrate/cps2001.html) 38 The migration of unmarried individuals, especially unmarried women, is driven by both marriage market and labor market considerations. For interesting discussions, see Edlund (2005) and Gautier, Svarer and Teulings (2010).
26
points. We argue that the mechanism through which proximity affects labor supply is the
availability of childcare. We interpret availability broadly enough to include not only
regular, scheduled childcare but also the insurance provided by the proximity of mothers
or mothers-in-law for irregular or unanticipated childcare needs.
Two endogeneity issues require attention. The first involves labor force participation
and childcare, and would arise even if proximity were exogenous. We address this by
using an IV approach, with proximity serving as an instrument for childcare. The second
is the potential endogeneity of proximity. Recent research on childcare and labor force
attachment has assumed exogeneity of proximity, suggesting that many economists view
this endogeneity as less serious than the first. It is also more difficult to address. We
address it by analyzing a subsample of military wives – civilian women with husbands
serving in the U.S. military – arguing that, compared with the general population, their
locations are more likely to be exogenous because their husbands' locations are primarily
determined by military needs.
Analysis of NSFH data suggests a strong relationship between proximity to mother or
mother-in-law and labor force attachment of married women with young children. Using
close proximity as an instrument for childcare hours, we find that both work-related and
non-work-related childcare by mothers or mothers-in-law increases the labor supply of
married women with young children. We interpret the significance of non-work-related
childcare as suggesting the insurance effect of proximity – the availability of family
members to provide irregular or unanticipated childcare increases the labor supply of
married women with young children.
We then turn to reduced form estimates of the relationship between labor supply and
proximity. We do this for two reasons. First, reduced form estimates for demographic
groups that do not benefit from the availability of childcare (e.g., married women without
young children) provide further evidence that proximity affects labor supply through the
availability of childcare. Second, the IV estimates using predicted childcare will
underestimate the insurance effect of close proximity if there are women for whom
childcare was available but not needed in the previous month.
We find that close proximity itself has a substantial, robust, and statistically
significant effect on labor force attachment for married women with young children. We
27
find no proximity effect for those demographic groups that would not benefit from the
availability of childcare: married women whose mothers or mothers-in-law are in poor
health, and women with only older children or no children. We find no proximity effect
for unmarried women with young children, a non-result we attribute to the inelastic labor
supply of unmarried women with children which makes them unresponsive to the
availability of childcare.
Census data provide further evidence. Using living in one's birth state as a proxy
for proximity to mother, we find that for married women with young children, birth state
residence increases the probability of labor force participation and employment by 4.0 -
6.1 percentage points. For married women without children and for never-married
women, we find a small, negative effect of living in the birth state of one or both spouses.
To control for the endogeneity of proximity, we consider the effects of proximity on
labor force attachment for a sample of military wives. We find that for military wives
with young children, living in the birth state of both spouses has a positive effect on labor
force attachment; we find no effect of birth state residence on military wives with young
children living only in his birth state or only in her birth state, and we find no effect of
birth state residence on military wives without young children.
This constellation of findings cannot be explained by either the network job
search hypothesis or by the tied mover hypothesis -- the proximity effects are too tightly
concentrated in a single demographic group -- married women with young children.
Because we have neither a natural experiment nor a structural model of proximity, we
cannot conclusively rule out selection. The military wives sample is as close as we come
to a natural experiment, but a skeptic could fairly point out that men self-select into the
military and women self-select into becoming and remaining military wives. Our
interstate movers sample is similarly open to the objection that individuals self-select into
return migration to his or her birth state.
The effects of close proximity on the labor supply of married women with young
children are substantial and robust. We find clear and convincing evidence that proximity
affects the labor force attachment of married women with young children, and that the
underlying mechanism is the availability of family members to meet childcare needs.
28
References
Arpino, Bruno, Chiara Pronzato and Lara Tavares, ``All in the Family: Informal Childcare and Mothers`Labour Market Participation.` ISER Working Paper Series, No. 2010-24, (July 9, 2010). Belan, Pascal, Pierre-Jean Messe, and François-Charles Wolff, "Postponing Retirement Age and Labor Force Participation: The Role of Family Transfers," Manuscript. March 2007. Bishop, Kelly, Bradley Heim, and Kata Mihaly, “Single Women’s Labor Supply Elasticities: Trends and Policy Implications,” Industrial and Labor Relations Review, Vol. 63, No. 1 (October 2009), 146-168. Blau, Francine D. and Lawrence M. Kahn, “Changes in the Labor Supply Behavior of Married Women: 1980-2000,” Journal of Labor Economics, Vol. 25, No. 3 (July 2007), 393-438. Cardia, Emanuela and Serena Ng, “Intergenerational Time Transfers and Childcare,” Review of Economic Dynamics, Vol. 6, No. 2, (April 2003), 431-454. Clark, William A.V. and Suzanne Davies Withers, “Disentangling the Interaction of Migration, Mobility and Labor-Force Participation,” Environment and Planning A, Vol. 34, No. 5, (May 2002), 923-945. Compton, Janice, "The Mom Effect: Family Proximity and the Labour Force Attachment of Women in Canada," Canadian Labour Market and Skills Researcher Network (CLSRN) Working Paper No. 87 (November 2011). Compton, Janice and Robert A. Pollak, "Why Are Power Couples Increasingly Concentrated in Large Metropolitan Areas?" Journal of Labor Economics, Vol. 25, No. 3, (July 2007), 475-512. Compton, Janice and Robert A. Pollak, "Proximity and Coresidence of Adult Children and their Parents: Description and Correlates," University of Michigan Retirement Research Center, Working Paper 2009-215, October 2009. Costa, Dora, "A House of Her Own: Old Age Assistance and the Living Arrangements of Older Nonmarried Women," Journal of Public Economics, Vol. 72, No. 1, (April 1999), 39-59. Costa, Dora L. and Matthew E. Kahn, "Power Couples: Changes in the Locational Choice of the College Educated, 1940-1990," Quarterly Journal of Economics, Vol. 115, No. 4, (November 2000), 1287-1315. Dimova, Ralitza and François-Charles Wolff, “Grandchild Care Transfers by Ageing Immigrants in France: Intra-household Allocation and Labour Market Implications,” European Journal of Population , Vol. 24, No. 3, (September, 2008), 315-340. Dimova, Ralitza and François-Charles Wolff, “Do Downward Private Transfers Enhance Maternal Labor Supply? Evidence from Around Europe,” Journal of Population Economics, Vol. 24, No. 3 (July, 2011), 911-933. Edlund, Lena, "Sex and the City," Scandinavian Journal of Economics, Vol. 107, No. 1, (March 2005), 25-44.
29
Gautier, Peter A., Michael Svarer, and Coen N. Teulings, "Marriage and the City: Search Frictions and the Sorting of Singles," Journal of Urban Economics, Vol. 67, No. 2, (March 2010), 206-218. Greene, William H. “Economics Courses in Liberal Arts Colleges: Further Results,” The Journal of Economic Education, Vol. 29, No. 4 (Autumn, 1998), 291-300. Greenwood, Michael J., “Human Migration: Theory, Models, and Empirical Studies,” Journal of Regional Science, Vol. 25, No. 4, (September 1985), 521-544. Jürges, Hendrik, “Gender Ideology, Division of Housework, and the Geographic Mobility of Families,” Review of Economics of the Household, Vol. 4, No. 4, (December 2006), 299-323. Kimmel, Jean, “Child Care Costs as a Barrier to Employment for Single and Married Mothers,” The Review of Economics and Statistics, Vol. 80, No. 2, (May 1998), 287-299. Klerman, Jacob Alex and Arleen Leibowitz, "Child Care and Women's Return to Work after Childbirth," American Economic Review, Vol. 80, No. 2, (May 1990), 284-288. Konrad, Kai, Harald Künemund, Kjell Erik Lommerud, and Julio R. Robledo, “Geography of the Family,” American Economic Review, Vol. 92, No. 4, (September 2002), 981-998. Laughlin, Lynda, “Who’s Minding the Kids? Child Care Arrangements: Spring 2005/Summer 2006.” Current Population Reports, P70-121. U.S. Census Bureau, Washington, DC, 2010. LeClere, Felicia B. and Diane K. McLaughlin, “Family Migration and Changes in Women’s Earnings: A Decomposition Analysis,” Population Research and Policy Review, Vol. 16, No. 4, (August 1997), 315-335. Lichter, Daniel T., “Socioeconomic Returns to Migration among Married Women,” Social Forces, Vol. 62, No. 2, (December 1983), 487-503. Lundberg, Shelly and Robert A. Pollak, "Efficiency in Marriage," Review of Economics of the Household, Vol. 1, No. 3, (September 2003), 153-168. Marr, William L. and Frank W. Millerd, “Migration and the Employment Status of Married Women,” Canadian Journal of Regional Science, Vol. 11, No. 1, (Spring 1988), 119-132. McKinnish, Terra, “Spousal Mobility and Earnings,” Demography, Vol. 45, No. 4, (November 2008), 829-849. Mincer, Jacob, “Family Migration Decisions,” Journal of Political Economy, Vol. 86, No. 5, (October 1978), 749-773. Mosisa, Abraham and Steven Hipple, “Trends in Labor Force Participation in the United States,” Monthly Labor Review, Vol. 129, No. 10, (October 2006), 35-57. Pezzin, Lilliana and Barbara Steinberg Schone, “Intergenerational Household Formation, Female Labor Supply and Informal Caregiving,” Journal of Human Resources, Vol. 34, No. 3, (Summer 1999), 475-503.
30
Rabe, Birgitta, “Dual-Earner Migration, Earnings Gains, Employment and Self-Selection,” Journal of Population Economic., Vol 24, No. 2 (April, 2011), 477-497. Rainer, Helmut and Thomas Siedler, “O Brother, Where Art Thou? The Effects of Having a Sibling on Geographic Mobility and Labor Market Outcomes,” Economica, Vol. 76, No. 303, (July 2009), 528-556. Ruggles, Steven, "The Decline of Intergenerational Coresidence in the United States, 1850 to 2000," American Sociological Review, Vol. 72, No. 6, (December 2007), 964-989 Ruggles, Steven, Matthew Sobek, Trent Alexander, Catherine A. Fitch, Ronald Goeken, Patricia Kelly Hall, Miriam King and Chad Ronnander, Integrated Public Use Microdata Series: version 4.0 [Machine-readable database], Minneapolis, MN: Minnesota Population Center [producer and distributor], 2009. Schwartz, Aba, “Interpreting the Effect of Distance on Migration,” Journal of Political Economy, Vol, 81, No. 5, (September/October 1973), 1153-1169. Sjaastad, Larry A., “The Costs and Returns of Human Migration.” Journal of Political Economy, Vol. 70, No. 5, pt. 2, (October 1962), 80-93. Shields, Michael P. and Gail M. Shields, “A Theoretical and Empirical Analysis of Family Migration and Household Production: U. S. 1980-1985,” Southern Economic Journal, Vol. 59, No. 4, (April 1993), 768-782. Spitze, Glenna, “The Effect of Family Migration on Wives Employment: How Long Does it Last?” Social Science Quarterly, Vol. 65, No.1, (March 1984), 21-36. Sweet, James A. and Larry L. Bumpass, The National Survey of Families and Households – Waves 1 and 2: Data Description and Documentation. Center for Demography and Ecology, University of Wisconsin-Madison (http://www.ssc.wisc.edu/nsfh/home.htm), 1996. Zamarro, Gema, “Family Labor Participation and Childcare Decisions: The Role of Grannies,” Unpublished Manuscript, RAND Corporation. (November 2009).
31
Table 1: Childcare Received by Proximity
NSFH Wave II. Weighted percentages. Sample includes all women with children 12 years and under, not coresiding with their mothers, over the age of 24, with mother and mother-in-law alive and living in the United States (ALUS). Respondents are asked whether they received work-related or non-work-related childcare in the past 12 months, and if yes, from whom. Work status is current. “Near” is 25 miles or less.
Work Related
Childcare
Not Work Related Childcare
Sample Size
MARRIED WOMEN From Her Mother Near Neither Mother 4.2 8.6 924 Near Only Hers 23.7 31.1 506 Near Only His 7.0 13.4 497 Near Both 26.8 36.8 1125 From His Mother
Near Neither Mother 2.7 7.4 924 Near Only Hers 2.9 5.7 506 Near Only His 19.4 25.4 497 Near Both 18.4 25.0 1125 UNMARRIED WOMEN From Her Mother Not Near Mother 13.5 14.7 144 Near Mother 28.6 47.7 274
32
Table 2: Married Women’s Labor Force Attachment by Proximity Does not
live near mother
Lives near mother
Coresides with mother
Non Married Women No Children 12 and under Does not Work 19.9 24.2 19.1 Works Part-time 16.8 15.6 21.6 Works Full-Time 63.3 60.3 59.3 Sample Size 303 348 92 Children 12 and under Does not Work 33.5 41.8 30.4 Works Part-time 16.0 9.6 23.4 Works Full-Time 50.5 48.5 35.2 Sample Size 144 274 46
Lives near
neither mother
Lives near his mother
only
Lives near her mother
only
Lives near
both mothers
Married Women No Children 12 and under Does not Work 23.3 28.6 22.0 20.8 Works Part-time 17.4 17.1 16.4 16.0 Works Full-Time 59.4 54.3 61.6 63.3 Sample Size 351 216 221 376 Children 12 and under Does not Work 43.9 33.7 41.1 35.9 Works Part-time 23.2 20.7 18.5 22.4 Works Full-Time 32.9 45.7 40.4 41.8 Sample Size 498 314 317 759 NSFH Wave II. Weighted percentages. Sample includes all women aged 25-60 whose mother is ALUS. The sample of married women includes only those for whom both mothers are ALUS. "Near" is 25 miles or less.
33
Table 3: Bivariate Probit Model: The Impact of Childcare on Positive Hours of Work
(A) Received work-related
childcare
(B) Received
non-work related childcare
(C) Received either work-
related or non-work related
childcare
Unmarried women with children 12 and under (367 observations) Childcare from her mother
Coefficient 0.068 -0.019 -0.007 Bootstrapped standard error (0.182) (0.073) (0.140)
Bootstrapped Marginal effect 0.024 -0.007 -0.002 Confidence interval (95%) (-0.318, 0.334) (-0.155, 0.312) (-0.322, 0.266)
Bootstrapped Wald coef 1.287 -0.133 -0.055
Married women with children 12 and under (1567 observations) Childcare from her mother
Coefficient 0.079 0.080 0.071 Bootstrapped standard error (0.052) (0.050) (0.042)
Bootstrapped Marginal effect -0.012 -0.007 -0.011 Confidence interval (95%) (-0.019, 0.178) (-0.025, 0.159) (-0.027, 0.143)
Bootstrapped Wald coef 1.565 1.604 1.646
Childcare from either mother Coefficient 0.165 0.165 0.136
Bootstrapped standard error (0.057) (0.061) (0.047) Marginal effect 0.062 0.062 0.051
Confidence interval (95%) (0.066, 0.273) (0.068, 0.319) (0.063, 0.253) Bootstrapped Wald coef 2.92 2.953 2.908
NSFH Wave II. The sample includes all married and unmarried women, aged 25-60 inclusive, with children 12 and under, for whom both mother (and mother-in-law if applicable) are ALUS.
34
Table 4: Probit and Tobit Regression Results Married Women Unmarried Women
Probit: Positive Hours of Work
Tobit: Usual Weekly Hours of Work
Probit: Positive Hours of
Work
Tobit: Usual
Weekly Hours of
Work (1) (2) (3) (4) (5) (6) Coreside with Mother --- --- --- --- -0.228* -4.103* (0.120) (2.203)
-0.073
Lives Near Own Mother
0.046 0.183 -0.096 -2.013 (0.075) (1.630) (0.106) (1.321)
---
---
Lives Near Own Mother Only
0.008 -1.076 (0.081) (1.752)
---
Lives Near Spouse's Mother Only
0.151* 3.519* (0.098) (2.045)
0.052
Lives Near Both Mothers
0.211** 3.429* (0.092) (1.897)
0.073
Pr(Y=1|X) 0.685 0.685 0.770 Y (fitted values) 19.32 19.29 28.14 Observations 2521 2513 2517 2509 1637 1637 Wald chi2 (DF)
407.33 (30)
380.05 (38)
1124.14 (30)
1947.90 (38)
190.8 (24)
243.0 (24)
Pseudo R2 0.082 0.087 0.017 0.018 0.081 0.013 NSFH Wave II. Coefficients are presented, with bootstrapped standard errors in parentheses. Marginal effects are in italics, presented if the coefficient is significant at the 80 percent confidence level. The sample includes all individuals in the marriage category, aged 25-60 inclusive, for whom both mother (and mother-in-law if applicable) are ALUS. The unmarried sample includes all individuals who are currently divorced, separated, widowed or never married. Control variables included in the regressions, but not presented here for space considerations include age, age squared, whether spouse works and his/her hours of work, whether self or spouse currently has medical problems, spouse’s income, race (Black, Hispanic, White (omitted)), education categories (both spouses have college degrees, only she has a college degree, only he has a college degree, neither has a college degree (omitted)), children 12 and under present in the household, only children over 12 present in the household, children outside the household, whether mother has a college degree, region (Midwest, South, West, Northeast (omitted)), average commuting time in the county, whether residing in an MSA, 1990 county level unemployment rate, whether coresides with mother or mother-in-law, age categories of mother(s) (less than 60, 60-69, 70 and over (omitted)), whether mother (or mother-in-law) is in poor health, whether siblings live within 25 miles.
35
Table 5: Probit Regressions, Dependent Variable: Positive Hours of Work Sample: Married Women, with both mother and mother-in-law ALUS
NSFH Wave II. Coefficients are presented, with standard errors in parentheses. Marginal effects on the predicted probability are italicized and listed for those coefficients that are statistically significant at the 80 percent confidence level. The sample includes all individuals in the marriage category, aged 25-60 inclusive, for whom both mother are ALUS. Full control variables are included in all regressions, see footnote from table 4 for the list of controls.
All Children 12 and
under in household
Mothers without young
children in household
Non-Mothers
Her mother in
poor health
His mother in
poor health
(1) (2) (3) (4) (5) (6) Lives near her mother only
0.008 0.097 -0.067 -0.093 -0.090 0.081 (0.081) (0.132) (0.228) (0.370) (0.406) (0.309)
---
--- --- --- --- ---
Lives near his mother only
0.151* 0.280*** -0.190 0.224 -0.034 0.004 (0.098) (0.094) (0.229) (0.308) (0.466) (0.316)
0.052
0.104 --- --- --- ---
Lives near both mothers
0.211** 0.266** 0.118 -0.020 0.085 0.077 (0.092) (0.110) (0.247) (0.407) (0.454) (0.372)
0.073
0.101 --- --- --- ---
Children 12 and under
-0.548*** -0.588* -0.547 (0.082) (0.311) (0.463)
-0.186 -0.188 -0.182 Her siblings within 25 miles
0.055 -0.014 0.063 0.572** 0.131 -0.069 (0.082) (0.102) (0.148) (0.266) (0.338) (0.377)
--- --- --- 0.147 --- --- His siblings within 25 miles
-0.161* -0.144 -0.082 -0.349 -0.181 0.121 (0.068) (0.091) (0.132) (0.375) (0.242) (0.310)
-0.057 -0.055 --- --- --- --- Pr(Y=1|X) 0.685 0.609 0.713 0.831 0.719 0.699 Observations 2,513 1564 589 356 318 286 Pseudo R2 0.087 0.066 0.139 0.188 0.165 -0.206
Table 6: Summary Statistics for Three U.S. Census Samples. Married Military Wives
Never Married
With Children12 and under
With Only Older
Children
No Children in
the Household
With Children 12 and under
With Only Older
Children
No Children in the
Household
With Children 12 and under
With Only Older
Children
No Children in the
Household
Sample Size 675,850 172,114 204.058 10.578 1,120 3,135 73,813 16,336 226,000 In Birth State 63.84% 66.96% 59.94% 18.95% 20.54% 19.40% 75.61% 74.54% 71.49% In Only Her Birth State 15.44% 14.11% 15.69% 10.24% 9.29% 10.85% In Only His Birth State 12.86% 12.61% 13.29% 5.08% 5.71% 5.39% In Both Birth States 48.40% 52.85% 44.25% 8.71% 11.25% 8.55% In the Labor Force Not Residing in Birth State 0.64 0.79 0.85 0.53 0.75 0.79 0.78 0.77 0.87 Residing in His Birth State 0.70 0.80 0.85 0.57 0.80 0.85 Residing in Her Birth State 0.69 0.80 0.85 0.60 0.81 0.84 0.74 0.72 0.81 Residing in Both Birth State 0.71 0.80 0.83 0.62 0.83 0.79 Employed Not Residing in Birth State 0.62 0.76 0.82 0.49 0.71 0.72 0.70 0.70 0.83 Residing in His Birth State 0.68 0.78 0.82 0.54 0.78 0.78 Residing in Her Birth State 0.66 0.77 0.83 0.56 0.79 0.76 0.65 0.65 0.76 Residing in Both Birth State 0.68 0.78 0.81 0.59 0.81 0.75 Usual Weekly Hours Not Residing in Birth State 25.13 31.6 37.2 22.50 30.34 34.59 32.44 32.89 37.76 Residing in His Birth State 27.08 32.0 36.3 24.11 30.25 34.96 Residing in Her Birth State 26.73 31.9 36.5 25.36 29.32 33.81 30.45 29.90 33.82 Residing in Both Birth State 26.91 31.4 35.1 25.55 31.93 34.28
U.S. Census 2000. The samples includes all married and single never married women aged 25-45, born in the U.S., non-students. The military wives sample includes all women aged 18-45, non-students, born in the U.S. whose husbands are employed in the U.S. military.
Table 7: Probit Regressions: Impact of Birth State on Labor Force Attachment MARRIED WOMEN MILITARY WIVES NEVER MARRIED
WOMEN In Labor
Force Employed In Labor
Force Employed In Labor
Force Employed
(1) (2) (3) (4) (5) (6) Living in birth state
-0.058*** -0.052*** -0.058 -0.026 -0.039** -0.041** (0.012) (0.012) (0.059) (0.026) (0.017) (0.016)
-0.008 -0.009 --- --- -0.008 -0.014 Children 12 and under
- 0.615***
-0.570***
-0.827***
-0.739***
0.012
-0.056**
(0.012) (0.012) (0.033) (0.032) (0.028) (0.027) -0.159 -0.155 -0.277 -0.266 -0.008** -0.011 Birth state X Children 12 and under
0.128***
0.122***
0.990
0.070
0.030
0.013
(0.014) (0.014) (0.066) (0.064) (0.032) (0.030) 0.039 0.039 0.032 0.026 --- ---
Prob(Y=1|X) 0.751 0.728 0.633 0.582 0.841 Observations 210004 210004 14833 14833 63182 63182 LRchi2 (DF)
17887.84 (33)
18428.96 (33)
1955.10 (33)
1968.20 (33)
9035.17 (28)
10939.58 (28)
Pseudo R2 0.074 0.073 0.099 0.097 0.146 0.155 U.S. Census 2000. Coefficients presented with standard error in parentheses and marginal effect on the predicted probability in parentheses. The sample includes all married and never married women aged 25-45, born in the U.S., non-students. The regressions use a random 10 percent sample. The military wives sample includes all women aged 18-45, non-students, born in the U.S. whose husbands are employed in the U.S. military. The full set of controls are included in each regression. These include age, age squared, children (children 12 and under, only children over 12 in the household, no children in the household (omitted)), education (less than high school, high school diploma (omitted), more than high school, bachelor’s degree, more than bachelor’s degree), spouse education (groups same), disability, spouse disability, race (Black, Hispanic, white (omitted)), rented accommodations, whether in a metropolitan area, total income of spouse, whether in different state five years prior, size of current state (square miles), U.S. region.
38
Table 8: Probit Regressions: Impact of Birth State on Labor Force Attachment MARRIED WOMEN MILITARY WIVES In Labor
Force Employed In Labor
Force Employed
(1) (2) (3) (4) Living in only her birth state
-0.045** -0.050*** 0.023 0.020 (0.019) (0.018) (0.077) (0.783)
-0.012
-0.011 --- ---
Living in only his birth state
-0.067*** -0.048** 0.080 0.036 (0.019) (0.019) (0.104) (0.709)
-0.011
-0.014 --- ---
Living in birth state of both
-0.088*** -0.069*** -0.140 -0.075 (0.015) (0.014) (0.080) (0.078)
-0.013
-0.016 --- ---
Children 12 and under
-0.668*** (0.015)
-0.131*** (0.021)
-0.821*** (0.034)
-0.735*** (0.033)
-0.158 -0.155 -0.277 -0.266 Her Birth State X Children 12 and under
0.128*** 0.131*** 0.018 0.018 (0.022) 0.040
(0.021) 0.042
(0.087) ---
(0.083) ---
His Birth State X Children 12 and under
0.160*** 0.141*** -0.102 -0.056 (0.023) 0.049
(0.022) 0.046
(0.119) ---
(0.113) ---
Both Birth State X Children 12 and under
0.202*** 0.184*** 0.176** 0.122 (0.017) 0.061
(0.016) 0.060
(0.091) 0.053
(0.088) 0.042
Prob(Y=1|X) 0.751 0.728 0.633 0.582 Observations 210004 210004 14833 14833 LRchi2 (DF)
17967.33 (37)
18499.93 (37)
1958.33 (37)
1969.41 (37)
Pseudo R2 0.074 0.074 0.099 0.097 U.S. Census 2000. Coefficients presented with standard error in parentheses and marginal effect on the predicted probability in parentheses. Sample and control variables are as described in table 8.
39
Table 9: Employment and Labor Force Participation by State of Residence 5 Years Prior Non-Migrants Migrants (A)
In Birth State
(B) Not in Birth State
(C) Difference for Non-Migrants (A) – (B)
(D) Into Birth
State
(E) Not Into
Birth State
(F) Difference
for Migrants (D) – (E)
Married with Children 12 and under Employed 0.683 0.671 0.013*** 0.581 0.535 0.054*** In Labor Force 0.704 0.689 0.016*** 0.608 0.552 0.055*** Married with Children over 12 Only Employed 0.780 0.778 0.002 0.688 0.712 -0.024** In Labor Force 0.801 0.799 0.001 0.723 0.749 -0.025** Married with no Children in Household Employed 0.815 0.825 -0.010*** 0.803 0.816 -0.013** In Labor Force 0.838 0.847 -0.009*** 0.837 0.849 -0.012* Single never married (SNM) with Children 12 and under Employed 0.652 0.696 -0.044*** 0.666 0.722 -0.056*** In Labor Force 0.743 0.775 -0.032*** 0.776 0.801 -0.025* SNM with Children over 12 only Employed 0.652 0.697 -0.045*** 0.622 0.724 -0.103** In Labor Force 0.721 0.761 -0.040*** 0.703 0.823 -0.121*** SNM with no Children in Household Employed 0.762 0.815 -0.053*** 0.807 0.869 -0.062*** In Labor Force 0.804 0.851 -0.046*** 0.858 0.905 -0.047*** U.S. Census 2000 unweighted IPUMS sample. Includes all women aged 25-45 and born in the U.S. The sample of migrants includes those who lived outside their birth state five years prior to the census, and have since migrated across states.
Table 10: Probit Regressions: Labor Force Participation Sample: Women who did not reside in their birth state five years prior to the census. MARRIED WOMEN NEVER MARRIED WOMEN
All
Children 12 and under
Only Children over 12
No Children in the
Household All
Children 12 and under
Only Children over 12
No Children in the
Household (1) (2) (3) (4) (5) (6) (7) (8) Omitted Case: (A) Non-migrant (B) Return migrant: Moved into birth state
-0.189*** -0.183*** -0.179*** -0.166*** 0.014 0.164* -0.353 -0.028 (0.021) (0.025) (0.069) (0.054) (0.042) (0.084) (0.236) (0.050)
-0.061 -0.066 -0.049 -0.036 --- 0.164 -0.100 --- (C) Onward Migrant: Moved into another state.
-0.291*** -0.332*** -0.201*** -0.200*** 0.012 -0.008 -0.153 0.011 (0.015) (0.018) (0.049) (0.036) (0.037) (0.084) (0.235) (0.042)
-0.095 -0.119 -0.055 -0.043 --- --- --- --- Children 12 and under -0.574*** -0.054*
(0.014) (0.031) -0.187 -0.011 Only Children over 12 in the household
-0.154*** 0.100* (0.019) (0.056)
-0.050 0.020 Chi2 : Test (C) = (E) 17.46 26.81 0.07 0.32 0.00 2.38 0.39 0.43 Prob>chi2 0.000 0.000 0.788 0.574 0.959 0.123 0.531 0511 Pr(Y=1|X) 0.739 0.676 0.806 0.866 0.880 0.790 0.796 0.901 Observations 74,036 47,592 11,233 15,211 19,306 3,520 796 14,990 Pseudo R2 0.080 0.633 0.645 0.103 0.158 0.078 0.103 0.180
U.S. Census 2000. Coefficients are presented with standard errors in parentheses, marginal effects in italics if p-value is less than 0.2. Sample: A twenty percent random sample of all women aged 25-45, born in the U.S. who were not living in their birth state five years prior to the census. The full set of controls, as described in table 8, is included in each regression.
Appendix 1: Means and Standard Deviations, NSFH Married Women Sample
Live near Neither Mother
Live Near His Mother
Only
Live near Her Mother
Only
Live near Both
Mothers Currently Working 0.648 0.627 0.689 0.693 (0.478) (0.484) (0.463) (0.461) Usual Weekly Hours (incl. 0) 23.893 22.993 26.577 25.261 (20.402) (19.693) (20.946) (19.484) Usual Weekly Hours (excl. 0) 36.906 36.690 38.642 36.166 (12.755) (10.783) (13.104) (7.746) Coreside with Her Mother 0.017 0.022 (0.130) (0.145) Coreside with His Mother 0.041 0.029 (0.198) (0.168) Different City Prior Wave 0.420 0.283 0.308 0.211 (0.494) (0.450) (0.462) (0.408) Her Siblings within 25 miles 0.145 0.709 0.333 0.820 (0.352) (0.454) (0.471) (0.385) His Siblings within 25 miles 0.165 0.319 0.725 0.824 (0.371) (0.466) (0.447) (0.381) Children 12 and under 0.546 0.562 0.559 0.635 (0.498) (0.496) (0.496) (0.481) Age 38.706 37.451 37.027 36.166 (8.124) (7.297) (7.586) (7.746) Medical Problems 0.205 0.246 0.245 0.197 (0.404) (0.431) (0.430) (0.398) Black 0.037 0.061 0.043 0.080 (0.188) (0.240) (0.203) (0.272) Hispanic 0.053 0.061 0.055 0.083 (0.223) (0.239) (0.229) (0.276) Mother has college degree 0.331 0.260 0.280 0.180 (0.471) (0.439) (0.449) (0.384) Half Power - He has College 0.202 0.115 0.111 0.103 (0.402) (0.319) (0.314) (0.303) Half Power - She has College 0.081 0.080 0.089 0.071 (0.273) (0.271) (0.284) (0.257) Power - Both have college 0.358 0.206 0.247 0.095 (0.480) (0.405) (0.431) (0.294) Spouse’s Income 43,442 36,394 33,619 32,632 (47,041) (32,548) (29,254) (23,819) He is not working 0.089 0.120 0.119 0.111 (0.285) (0.325) (0.323) (0.314) His usual weekly hours (incl. 0) 42.522 41.571 41.148 41.706 (16.577) (18.765) (18.372) (18.214) He has Medical Problems 0.246 0.262 0.217 0.289 (0.431) (0.440) (0.412) (0.453)
…
42
Live near Neither Mother
Live Near His Mother
Only
Live near Her Mother
Only
Live near Both
Mothers Midwest 0.264 0.221 0.352 0.275 (0.441) (0.415) (0.478) (0.447) South 0.361 0.337 0.302 0.359 (0.480) (0.473) (0.459) (0.480) West 0.249 0.249 0.197 0.145 (0.433) (0.433) (0.398) (0.352) Average Commuting Time 22.101 22.266 21.928 21.780 (4.040) (3.812) (4.554) (4.100) Resides in MSA 0.856 0.824 0.819 0.766 (0.351) (0.381) (0.385) (0.423) MSA Unemployment Rate 5.970 6.353 6.145 6.637 (1.678) (1.742) (1.866) (2.248) Her Mother Aged Less than 60 0.291 0.305 0.344 0.377 (0.454) (0.460) (0.475) (0.485) Her Mother Aged 60-69 0.332 0.345 0.354 0.342 (0.471) (0.475) (0.478) (0.474) Her Mother Widowed/Divorced 0.266 0.305 0.218 0.257 (0.442) (0.460) (0.413) (0.437) Her Mother in Poor Health 0.142 0.132 0.122 0.134 (0.349) (0.338) (0.328) (0.340) His Mother Aged Less than 60 0.206 0.229 0.319 0.320 (0.404) (0.420) (0.466) (0.466) His Mother Aged 60-69 0.349 0.350 0.336 0.340 (0.477) (0.477) (0.472) (0.474) His Mother Widowed/Divorced 0.291 0.311 0.273 0.298 (0.454) (0.463) (0.446) (0.457) His Mother in Poor Health 0.102 0.168 0.093 0.120 (0.303) (0.373) (0.291) (0.325) NSFH Wave II. Weighted percentages. Sample includes all married women aged 25-60, non-students, whose mother and mother-in-law is ALUS. "Near" is 25 miles or less.
43
Appendix 2: Means and Standard Deviations, 2000 Census Data, Married Women Sample
Live in birth state of neither spouse
Live in birth state of husband
only
Live in birth state
of wife only
Live in birth state
of both spouses
Employed 0.685 0.719 0.714 0.723
(-0.465) (-0.450) (-0.452) (-0.448)
In the Labor Force 0.708 0.739 0.737 0.744
(-0.455) (-0.439) (-0.440) (-0.436)
Age 36.636 36.254 36.103 36.177
(-5.590) (-5.662) (-5.717) (-5.705)
Less than High School 0.064 0.076 0.086 0.092
(-0.245) (-0.266) (-0.281) (-0.289)
More than High School 0.335 0.353 0.346 0.337
(-0.472) (-0.478) (-0.476) (-0.473)
Bachelors 0.259 0.219 0.205 0.160
(-0.438) (-0.414) (-0.403) (-0.367)
More than Bachelor 0.117 0.080 0.080 0.055
(-0.321) (-0.271) (-0.271) (-0.229)
Spouse less than high school 0.080 0.097 0.119 0.119
(-0.272) (-0.296) (-0.324) (-0.324)
Spouse more than high school 0.299 0.319 0.318 0.305
(-0.458) (-0.466) (-0.466) (-0.460)
Spouse bachelors 0.247 0.197 0.194 0.148
(-0.431) (-0.398) (-0.395) (-0.356)
Spouse more than bachelors 0.165 0.089 0.101 0.055
(-0.371) (-0.285) (-0.301) (-0.228)
Spouse work disability 0.084 0.097 0.096 0.102
(-0.277) (-0.296) (-0.294) (-0.302)
Work disability 0.068 0.078 0.075 0.078
(-0.253) (-0.268) (-0.264) (-0.269)
Black 0.060 0.053 0.063 0.071
(-0.237) (-0.224) (-0.243) (-0.257)
Hispanic 0.043 0.033 0.081 0.047
(-0.203) (-0.179) (-0.273) (-0.212)
Renter (free rent) 0.023 0.017 0.015 0.016
(-0.150) (-0.129) (-0.121) (-0.127)
Renter 0.198 0.164 0.185 0.147
(-0.398) (-0.370) (-0.389) (-0.354)
44
Live in birth state of neither spouse
Live in birth state of husband
only
Live in birth state
of wife only
Live in birth state
of both spouses
Metropolitan Area 0.745 0.639 0.697 0.616
(-0.436) (-0.480) (-0.460) (-0.486)
Spouse's total income 60,539 50,786 51,392 45,894
(-65,287) (-54,755) (-55,358) (-46,326)
Migrant 0.292 0.095 0.069 0.023
(-0.455) (-0.293) (-0.254) (-0.149)
Children less than 12 0.641 0.640 0.649 0.644
(-0.480) (-0.480) (-0.477) (-0.479)
Children over 12 0.142 0.158 0.151 0.178
(-0.349) (-0.365) (-0.358) (-0.383)
No Children 0.218 0.202 0.200 0.178
(-0.413) (-0.401) (-0.400) (-0.382)
Mid-Atlantic region 0.071 0.109 0.116 0.180
(-0.256) (-0.312) (-0.320) (-0.384)
East North Central region 0.292 0.181 0.173 0.128
(-0.455) (-0.385) (-0.378) (-0.334)
West North Central region 0.061 0.079 0.068 0.074
(-0.240) (-0.270) (-0.251) (-0.263)
South Atlantic Region 0.101 0.126 0.129 0.116
(-0.302) (-0.332) (-0.335) (-0.320)
East South Central Region 0.094 0.168 0.163 0.236
(-0.291) (-0.374) (-0.369) (-0.425)
West South Central Region 0.060 0.094 0.077 0.092
(-0.238) (-0.292) (-0.266) (-0.289)
Mountain 0.135 0.070 0.065 0.032
(-0.342) (-0.255) (-0.247) (-0.176)
Pacific 0.131 0.121 0.160 0.092 (-0.338) (-0.327) (-0.367) (-0.289)
US Census 2000. Sample includes all married women aged 25-45, non-students.
45
Appendix 3: Means and Standard Deviations, 2000 Census Data, Military Wives Sample
Live in birth state of neither spouse
Live in birth state of husband
only
Live in birth state
of wife only
Live in birth state
of both spouses
Employed 0.556 0.613 0.621 0.643
(-0.497) (-0.487) (-0.485) (-0.479)
In the Labor Force 0.603 0.653 0.663 0.675
(-0.489) (-0.476) (-0.473) (-0.469)
Age 31.398 32.316 31.046 31.873
(-6.710) (-6.703) (-6.848) (-6.940)
Less than High School 0.046 0.035 0.042 0.062
(-0.209) (-0.184) (-0.200) (-0.241)
More than High School 0.426 0.473 0.481 0.408
(-0.494) (-0.500) (-0.500) (-0.492)
Bachelors 0.202 0.191 0.167 0.147
(-0.401) (-0.393) (-0.373) (-0.354)
More than Bachelor 0.057 0.061 0.040 0.049
(-0.231) (-0.240) (-0.196) (-0.215)
Spouse less than high school 0.010 0.013 0.008 0.022
(-0.097) (-0.113) (-0.088) (-0.147)
Spouse more than high school 0.471 0.482 0.516 0.505
(-0.499) (-0.500) (-0.500) (-0.500)
Spouse bachelors 0.165 0.179 0.139 0.138
(-0.371) (-0.384) (-0.346) (-0.345)
Spouse more than bachelors 0.134 0.105 0.081 0.055
(-0.341) (-0.307) (-0.272) (-0.228)
Spouse work disability 0.060 0.045 0.062 0.091
(-0.237) (-0.208) (-0.242) (-0.288)
Work disability 0.047 0.065 0.053 0.062
(-0.211) (-0.247) (-0.224) (-0.242)
Black 0.125 0.118 0.116 0.129
(-0.330) (-0.323) (-0.320) (-0.336)
Hispanic 0.063 0.062 0.092 0.111
(-0.243) (-0.242) (-0.290) (-0.314)
Renter (free rent) 0.299 0.178 0.188 0.133
(-0.458) (-0.383) (-0.391) (-0.340)
Renter 0.364 0.294 0.348 0.272
(-0.481) (-0.456) (-0.476) (-0.445)
46
Live in birth state of neither spouse
Live in birth state of husband
only
Live in birth state
of wife only
Live in birth state
of both spouses
Metropolitan Area 0.771 0.778 0.772 0.673
(-0.420) (-0.416) (-0.420) (-0.469)
Spouse's total income 36,979 37,746 35,162 34,950
(-23,823) (-18,813) (-20,921) (-19,832)
Migrant 0.756 0.521 0.312 0.230
(-0.430) (-0.500) (-0.464) (-0.421)
Children less than 12 0.716 0.697 0.709 0.700
(-0.451) (-0.460) (-0.454) (-0.458)
Children over 12 0.074 0.083 0.068 0.096
(-0.261) (-0.276) (-0.252) (-0.294)
No Children 0.210 0.219 0.223 0.204
(-0.407) (-0.414) (-0.416) (-0.403)
Mid-Atlantic region 0.035 0.071 0.045 0.092
(-0.184) (-0.258) (-0.208) (-0.289)
East North Central region 0.414 0.323 0.291 0.219
(-0.493) (-0.468) (-0.454) (-0.414)
West North Central region 0.074 0.058 0.048 0.083
(-0.262) (-0.235) (-0.215) (-0.276)
South Atlantic Region 0.108 0.144 0.163 0.134
(-0.310) (-0.351) (-0.370) (-0.341)
East South Central Region 0.026 0.065 0.048 0.112
(-0.158) (-0.247) (-0.213) (-0.315)
West South Central Region 0.045 0.036 0.043 0.079
(-0.208) (-0.187) (-0.202) (-0.270)
Mountain 0.087 0.056 0.063 0.041
(-0.282) (-0.230) (-0.243) (-0.199)
Pacific 0.189 0.208 0.270 0.203 (-0.392) (-0.406) (-0.444) (-0.402)
US Census 2000. Sample includes all civilian women married to active federal military employees, aged 18-45, non-students.